Image sensors typically are in the form of linear and area image sensors. These sensors are often provided by charge coupled devices (CCDs). Area image sensors can take the form of interline image sensors and full frame image sensors. There are a number of defects which affect image quality in image sensors. Of particular concern is a hopping pixel defect. This defect often comes up in a random fashion and so unless it is detected in an image sensor through manufacturing quality control, the image sensor will be delivered to the general public and have this defect. A hopping pixel defect is defined as a pixel whose dark signal level varies beyond the random noise along with the time.
FIG. 1a depicts a series of test for an individual pixel in an image sensor in which the output from the series of test is plotted as signal level versus time. FIG. 1a also illustrates a representative output for a particular pixel that does not have a hopping pixel defect. FIG. 1b depicts a histogram for the data of the pixel shown in FIG. 1a which is a plot of the counts or frequency versus signal level. It shows that the normal dark signal obeys a Gaussian distribution and it has only one peak which means the pixel has only one signal level. FIGS. 1c and 1d show similar plots to those respectively in FIGS. 1a and 1b, but in this case an individual pixel has a hopping pixel defect. In FIG. 1d the hopping pixel has two peaks which represent two different signal levels. One is the normal dark signal level (about 372 counts) and the other one is the hopping signal level (about 386 counts). In this case, 8 digital counts equal to 1 millivolt. Therefore, the hopping magnitude is about 1.75 millivolt.
Heretofore it has been difficult to detect such defects. The difficulty of the task is that the hopping pixel signal is very small (almost down to a couple of millivolts, close to the test system noise level) and hopping rate is random (some defects may hop once per milliseconds and others may take minutes or even hours to hop once).